Let’s Talk About AI and Its Carbon Footprint

Let’s take a moment to chat about the less glamorous side of artificial intelligence. We like to think of AI as this magical, invisible thing, just lines of code floating around in “the cloud.” But here’s the truth: the cloud isn’t some dreamy, weightless place. It’s very real, very physical, and very energy-hungry. Behind every clever chatbot like ChatGPT, every voice assistant on your phone like Siri, and every curated Spotify playlist, there are banks of servers, massive data centres, industrial cooling systems and a whole lot of electricity keeping it all running.

How Bad Is It?

You know how boiling a kettle feels like a waste when you only need a drop? Well, training a large AI model like GPT or DALL·E isn’t just boiling one kettle, it’s boiling a stadium full of kettles, repeatedly, for weeks.

Let us give you a rough idea: training a big AI model can emit hundreds of tonnes of CO₂. Some estimates say the carbon cost of training a single large model can equal the lifetime emissions of five cars and that’s just training. Once it’s trained, you have inference, the bit where it answers your question about why sourdough isn’t rising or recommends a recipe for dinner. That happens millions of times a day, which adds up too.

Where’s All That Energy Going?

A big chunk goes into running high-performance computers 24/7. These aren’t your average laptops, they’re machines packed into enormous warehouses, often in the US, Ireland, or Scandinavia and they need to stay cool (because otherwise they literally overheat), so you have air-conditioning systems, fans, and sometimes even water-cooled floors.

And guess what powers most of this? In many places, fossil fuels still do the heavy lifting.

So What’s Being Done?

The tech world is kind of waking up. Companies are:

• Using renewable energy where they can.

• Trying to make models more efficient, needing fewer resources to train.

• Sharing research on low-impact AI.

• Investing in carbon offsets, though that’s a bit like buying a cake and planting a tree to cancel it out, it’s helpful, but doesn’t tackle the root.

Plus, some newer AI models are being designed to be lighter and leaner, delivering decent performance without guzzling energy.

What Can We Do?

• Be mindful of how you use AI. You don’t need to ask it what to have for dinner. there are loads of brilliant cookbooks full of tasty, sustainable recipes that are better for the planet and a lot more fun to flick through. As for the simpler stuff? Maybe skip the chatbot and ask a human instead. A friend, your mum, your neighbour, they might just have the perfect answer (and you'll probably get a good story too).

• Support companies pushing for sustainable AI. It’s not just about using less energy, it’s about transparency, accountability, and investing in greener solutions from the ground up. Look out for tech companies that power their data centres with renewables, design more efficient systems, and openly share their progress. Whether it’s Microsoft teaming up with Irish wind farms, Google working toward carbon-free energy 24/7, or smaller platforms building AI tools that don’t drain the grid, where we put our support (and our clicks) really does matter.

• Talk about it, because awareness matters and this kind of thing doesn’t usually make the front page. We hear plenty about AI breakthroughs and shiny new tech, but rarely about the environmental cost behind it or the companies quietly trying to do better. The more we talk about the impact, the more pressure there is on big players to step up.

AI is powerful, and when used wisely, it can help solve climate problems (smart grids, optimised farming, energy forecasting etc.). But if we’re not careful, we risk solving problems with one hand while quietly making new ones with the other.

So next time you fire up ChatGPT or ask your phone what sound a monkey makes, just remember: there’s a little environmental cost behind that convenience. Doesn’t mean we stop using it but it does mean we need to think smarter, and demand better from the companies building our digital future.